Update functions.py

functions.py

@@ -31,8 +31,10 @@ from langchain.vectorstores import FAISS
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain.chat_models import ChatOpenAI
 from langchain.callbacks import StdOutCallbackHandler
-from langchain.chains import ConversationalRetrievalChain, QAGenerationChain, RetrievalQA
+from langchain.chains import ConversationalRetrievalChain, QAGenerationChain, LLMChain
 from langchain.memory import ConversationBufferMemory
+from langchain.chains.question_answering import load_qa_chain
+from langchain.chains.conversational_retrieval.prompts import CONDENSE_QUESTION_PROMPT
 
 from langchain.prompts.chat import (
     ChatPromptTemplate,
@@ -87,7 +89,11 @@ def load_prompt():
     ----------------
     {context}"""
     
-    prompt = system_template
+    messages = [
+        SystemMessagePromptTemplate.from_template(system_template),
+        HumanMessagePromptTemplate.from_template("{question}")
+    ]
+    prompt = ChatPromptTemplate.from_messages(messages)
         
     return prompt
 
@@ -566,13 +572,14 @@ def embed_text(query,_docsearch):
     #                              retriever=_docsearch.as_retriever(), 
     #                              return_source_documents=True)
     
-    chain = ConversationalRetrievalChain.from_llm(chat_llm, 
-                                                  retriever= _docsearch.as_retriever(search_kwargs={"k": 3}),
-                                                  get_chat_history=lambda h : h,
-                                                  memory = memory,
-                                                  return_source_documents=True)
-
-    chain.combine_docs_chain.llm_chain.prompt.messages[0] = SystemMessagePromptTemplate.from_template(load_prompt())
+    question_generator = LLMChain(llm=chat_llm, prompt=CONDENSE_QUESTION_PROMPT) 
+    doc_chain = load_qa_chain(llm=chat_llm,chain_type="stuff",prompt=load_prompt()) 
+    chain = ConversationalRetrievalChain(retriever=cfa_db.as_retriever(search_kwags={"k": 3}), 
+                                     question_generator=question_generator, 
+                                     combine_docs_chain=doc_chain, 
+                                     memory=memory, 
+                                     return_source_documents=True, 
+                                     get_chat_history=lambda h :h)
     
     answer = chain({"question": query})